Cooke Cd

Analyzing and Predicting Images Through a Neural Network Approach

A neural network approach has been developed to predict diagnostic image information to assist in the assessment of coronary artery disease. The predicted information represents the redistribution (or reversibility) of perfusion in the myocardium. A multilayer, backpropagation neural network is trained to predict the redistribution information from two other types of images: stress perfusion and myocardial thickening using SPECT imaging. The significance of this approach is two-fold: (i) the predicted reversibility information obviates the additional acquisition of delayed images (with the patient at rest), and (ii) the neural network approach represents a novel way with which to analyze and predict images from other images. This paper presents the methods that underlie the approach, and discusses the most recent experimental results that demonstrate its viability.

Automatic unification of three-dimensional cardiac perfusion with three-dimensional coronary artery anatomy

A method for automatic unification of three-dimensional models of the coronary artery tree from biplane angiograms and three-dimensional models of the left ventricular epicardial surface from perfusion SPECT is introduced. The efficacy of the technique is evaluated using six swine models. The results are shown to be satisfactory in all cases.

3-dimensional registration and visualization of reconstructed coronary arterial trees on myocardial perfusion distributions

Models created independently from nuclear perfusion images of the myocardium and x-ray angiograms of the coronary arteries are placed in register to produce a unified three- dimensional representation that more clearly displays and quantifies the relationship between stenotic defects in the arterial tree and perfusion distribution in the myocardium

An integrated processing program for analysis of left ventricular function and perfusion

A processing program for analysis of left ventricular perfusion, anatomy, and function from single photon emission computed tomographic perfusion images is presented. This program combines a well-known and validated perfusion quantification approach with a left ventricular surface modeler which enables computation of local and global functional parameters. Thus, perfusion and function can be associated and compared. The results can be displayed in two or three dimensions, using slice displays, polar plots, or surface-shaded 3-D graphics. Finally, the abilities to align a 3-D coronary artery tree with the 3-D left ventricular epicardial surface and display the unification results have been provided, to allow for comparison between coronary anatomy and left ventricular structure and function. Each portion of the analysis package has been separately validated.

Optimizing gated myocardial perfusion imaging processing for phase analysis.

In recent years, analysis of tomographic gated myocardial perfusion imaging (gMPI) has become increasingly sophisticated, and now includes measurement of left ventricle (LV) phase and dyssynchrony. Measurement of dyssynchrony has been suggested to have long-term prognostic value, and as a better way to identify clinically important multivessel coronary artery disease. Characterizing the magnitude of dyssynchrony as well as the area of latest contraction may contribute to the patient’s referral to, or exclusion from, high-cost interventions such as cardiac resynchronization therapy (CRT). Careful determination of processing parameters is critically important, to avoid artifacts and ensure consistent and clinically meaningful results. Parameters (apex, base, center, radial extent) which are acceptable for functional analysis, are not necessarily optimal for phase analysis. Moreover, phase analysis is important in some patient groups not historically referred for gMPI, whose studies can present challenges for processing. In addition to perfusion defects, these patients may have significant heart failure, and have very poor function or large, remodeled volumes. In most publications which incorporate phase analysis results, the approach used for verifying processing parameters is not emphasized. A systematic approach to processing for obtaining optimal phase analysis is needed, and is the subject of this article.

PERFUSE: an interactive knowledge-based system for the interpretation and explanation of cardiac imagery

The clinical evaluation of nuclear medicine myocardial perfusion studies is an information-intensive task requiring accuracy, consistency, and speed. Our aim is to facilitate this task by creating a knowledge-based approach assisting clinicians in their image-interpretation work. The approach is based on the design, implementation, and testing of a clinically useful, interactive, knowledge-based system to assist in the interpretation of myocardial perfusion distributions. This system extracts and transforms patient data into symbolic representations, capturing the visual reasoning of experts and thus providing a computerized assessment of the study. This assessment is presented as both a textual report and in graphical form. The review process itself is supported through an interactive (English-language text) report justification and explanation process. This paper focuses on the creation of the graphical user interface, and on the generation of the textual information derived from the numerical system output, both of which have been combined into a comprehensive, dynamic working environment.